Credit Card Usage Segmentation

This project focuses on developing unsupervised learning models for customer segmentation based on credit card usage data. The models aim to provide insights into distinct customer segments, improve credit risk assessment, and optimize marketing strategies. Deliverables include segmentation models, comprehensive data visualizations, and a comparative analysis of various algorithms.

Import necessary libraries

#Libraries for Data Analysis
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px


#for preprocessing, clustering, PCA
from sklearn.preprocessing import MinMaxScaler, StandardScaler, MaxAbsScaler, RobustScaler
from sklearn.cluster import KMeans, DBSCAN, AgglomerativeClustering
from sklearn.metrics import silhouette_score
from sklearn.decomposition import PCA
import warnings
warnings.filterwarnings('ignore')

#load the data
from google.colab import drive
drive.mount('/content/drive')

import os
os.getcwd()
os.chdir('/content/drive/MyDrive/DATA SCIENCE PROJECT/Credit Card Usage Segmentation')

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).

Load the dataset and review the features to understand their distributions, ranges, and relationships.

#import Customer Data.csv
df = pd.read_csv('Customer Data.csv', index_col = 'CUST_ID')

#Data Exploration
df.shape

(8950, 17)

df.head()

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE
CUST_ID
C10001	40.900749	0.818182	95.40	0.00	95.4	0.000000	0.166667	0.000000	0.083333	0.000000	0	2	1000.0	201.802084	139.509787	0.000000	12
C10002	3202.467416	0.909091	0.00	0.00	0.0	6442.945483	0.000000	0.000000	0.000000	0.250000	4	0	7000.0	4103.032597	1072.340217	0.222222	12
C10003	2495.148862	1.000000	773.17	773.17	0.0	0.000000	1.000000	1.000000	0.000000	0.000000	0	12	7500.0	622.066742	627.284787	0.000000	12
C10004	1666.670542	0.636364	1499.00	1499.00	0.0	205.788017	0.083333	0.083333	0.000000	0.083333	1	1	7500.0	0.000000	NaN	0.000000	12
C10005	817.714335	1.000000	16.00	16.00	0.0	0.000000	0.083333	0.083333	0.000000	0.000000	0	1	1200.0	678.334763	244.791237	0.000000	12

df.describe()

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE
count	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8950.000000	8949.000000	8950.000000	8637.000000	8950.000000	8950.000000
mean	1564.474828	0.877271	1003.204834	592.437371	411.067645	978.871112	0.490351	0.202458	0.364437	0.135144	3.248827	14.709832	4494.449450	1733.143852	864.206542	0.153715	11.517318
std	2081.531879	0.236904	2136.634782	1659.887917	904.338115	2097.163877	0.401371	0.298336	0.397448	0.200121	6.824647	24.857649	3638.815725	2895.063757	2372.446607	0.292499	1.338331
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	50.000000	0.000000	0.019163	0.000000	6.000000
25%	128.281915	0.888889	39.635000	0.000000	0.000000	0.000000	0.083333	0.000000	0.000000	0.000000	0.000000	1.000000	1600.000000	383.276166	169.123707	0.000000	12.000000
50%	873.385231	1.000000	361.280000	38.000000	89.000000	0.000000	0.500000	0.083333	0.166667	0.000000	0.000000	7.000000	3000.000000	856.901546	312.343947	0.000000	12.000000
75%	2054.140036	1.000000	1110.130000	577.405000	468.637500	1113.821139	0.916667	0.300000	0.750000	0.222222	4.000000	17.000000	6500.000000	1901.134317	825.485459	0.142857	12.000000
max	19043.138560	1.000000	49039.570000	40761.250000	22500.000000	47137.211760	1.000000	1.000000	1.000000	1.500000	123.000000	358.000000	30000.000000	50721.483360	76406.207520	1.000000	12.000000

df.info()

<class 'pandas.core.frame.DataFrame'>
Index: 8950 entries, C10001 to C19190
Data columns (total 17 columns):
 #   Column                            Non-Null Count  Dtype  
---  ------                            --------------  -----  
 0   BALANCE                           8950 non-null   float64
 1   BALANCE_FREQUENCY                 8950 non-null   float64
 2   PURCHASES                         8950 non-null   float64
 3   ONEOFF_PURCHASES                  8950 non-null   float64
 4   INSTALLMENTS_PURCHASES            8950 non-null   float64
 5   CASH_ADVANCE                      8950 non-null   float64
 6   PURCHASES_FREQUENCY               8950 non-null   float64
 7   ONEOFF_PURCHASES_FREQUENCY        8950 non-null   float64
 8   PURCHASES_INSTALLMENTS_FREQUENCY  8950 non-null   float64
 9   CASH_ADVANCE_FREQUENCY            8950 non-null   float64
 10  CASH_ADVANCE_TRX                  8950 non-null   int64  
 11  PURCHASES_TRX                     8950 non-null   int64  
 12  CREDIT_LIMIT                      8949 non-null   float64
 13  PAYMENTS                          8950 non-null   float64
 14  MINIMUM_PAYMENTS                  8637 non-null   float64
 15  PRC_FULL_PAYMENT                  8950 non-null   float64
 16  TENURE                            8950 non-null   int64  
dtypes: float64(14), int64(3)
memory usage: 1.2+ MB

Check for missing values, outliers, and any anomalies.

df.isnull().sum() #Check for missing values

	0
BALANCE	0
BALANCE_FREQUENCY	0
PURCHASES	0
ONEOFF_PURCHASES	0
INSTALLMENTS_PURCHASES	0
CASH_ADVANCE	0
PURCHASES_FREQUENCY	0
ONEOFF_PURCHASES_FREQUENCY	0
PURCHASES_INSTALLMENTS_FREQUENCY	0
CASH_ADVANCE_FREQUENCY	0
CASH_ADVANCE_TRX	0
PURCHASES_TRX	0
CREDIT_LIMIT	1
PAYMENTS	0
MINIMUM_PAYMENTS	313
PRC_FULL_PAYMENT	0
TENURE	0

dtype: int64

There's 1 missing value in Credit Limit and 313 missing values in Minimum Payments

df.duplicated().sum() #Check for duplicates

0

There's no duplicates in the dataset.

Handle the missing values.

The missing value in CREDIT_LIMIT can be droped since it's only 1.

df.dropna(subset = ['CREDIT_LIMIT'], inplace = True) #Drop Credit limit missing value since there's only 1 missing value

For the missing values in MINIMUM_PAYMENTS:

df[['PAYMENTS', 'MINIMUM_PAYMENTS']][df['MINIMUM_PAYMENTS'].isna()] #display PAYMENTS and MINIMUM_PAYMENTS

	PAYMENTS	MINIMUM_PAYMENTS
CUST_ID
C10004	0.000000	NaN
C10047	0.000000	NaN
C10049	0.000000	NaN
C10056	417.016763	NaN
C10057	0.000000	NaN
...	...	...
C19160	0.000000	NaN
C19170	0.000000	NaN
C19176	0.000000	NaN
C19185	0.000000	NaN
C19187	275.861322	NaN

313 rows × 2 columns

Since customers that don't have payments won't have minimum payment entry, if PAYMENTS = 0 then MINIMUM PAYMENTS = 0

#look for missing values with PAYMENTS above MINIMUM_PAYMENTS

print('Shape:', df[['PAYMENTS', 'MINIMUM_PAYMENTS']][(df['MINIMUM_PAYMENTS'].notna()) & (df['PAYMENTS'] > df['MINIMUM_PAYMENTS'])].shape)
df[['PAYMENTS', 'MINIMUM_PAYMENTS']][(df['MINIMUM_PAYMENTS'].notna()) & (df['PAYMENTS'] > df['MINIMUM_PAYMENTS'])].head()

Shape: (6272, 2)

	PAYMENTS	MINIMUM_PAYMENTS
CUST_ID
C10001	201.802084	139.509787
C10002	4103.032597	1072.340217
C10005	678.334763	244.791237
C10007	6354.314328	198.065894
C10008	679.065082	532.033990

#look for missing values with PAYMENTS below MINIMUM_PAYMENTS

print('Shape:', df[['PAYMENTS', 'MINIMUM_PAYMENTS']][(df['MINIMUM_PAYMENTS'].notna()) & (df['PAYMENTS'] < df['MINIMUM_PAYMENTS'])].shape)
df[['PAYMENTS', 'MINIMUM_PAYMENTS']][(df['MINIMUM_PAYMENTS'].notna()) & (df['PAYMENTS'] < df['MINIMUM_PAYMENTS'])].head()

Shape: (2364, 2)

	PAYMENTS	MINIMUM_PAYMENTS
CUST_ID
C10003	622.066742	627.284787
C10006	1400.057770	2407.246035
C10011	1083.301007	2172.697765
C10015	805.647974	989.962866
C10016	1993.439277	2109.906490

Payments should be completed if PAYMENTS is greater than MINIMUM_PAYMENT 2364 customers failed to paid below MINIMUM_PAYMENT which lead to invalid values

Therefore, fill the missing values by the mean of PAYMENTS if PAYMENTS is less than MINIMUM_PAYMENTS, the missing values will be filled by the correspond PAYMENTS

minimum_copy = df['MINIMUM_PAYMENTS'].copy() # make a copy of MINIMUM_PAYMENTS
payments_mean = np.mean(df['PAYMENTS']) # take the mean value of PAYMENTS

i = 0
for payments, minpayments in zip(df['PAYMENTS'], df['MINIMUM_PAYMENTS'].isna()):
    if (payments == 0) and (minpayments == True):
        minimum_copy[i] = 0
    elif (0 < payments < payments_mean) and (minpayments == True):
        minimum_copy[i] = payments
    elif minpayments == True:
        minimum_copy[i] = payments_mean
    i += 1

df['MINIMUM_PAYMENTS'] = minimum_copy.copy()

#final check of missing values
df.isnull().sum()

	0
BALANCE	0
BALANCE_FREQUENCY	0
PURCHASES	0
ONEOFF_PURCHASES	0
INSTALLMENTS_PURCHASES	0
CASH_ADVANCE	0
PURCHASES_FREQUENCY	0
ONEOFF_PURCHASES_FREQUENCY	0
PURCHASES_INSTALLMENTS_FREQUENCY	0
CASH_ADVANCE_FREQUENCY	0
CASH_ADVANCE_TRX	0
PURCHASES_TRX	0
CREDIT_LIMIT	0
PAYMENTS	0
MINIMUM_PAYMENTS	0
PRC_FULL_PAYMENT	0
TENURE	0

dtype: int64

Feature Scaling ( log scaler & MinMax Scaler )

Log transformation reduces skewness and compresses extreme values, but it doesn’t standardize the data to a particular range. Applying Min-Max scaling afterward can bring all features into a common range (typically 0 to 1), making them easier to interpret and ensuring they contribute equally to distance calculations in algorithms like K-Means.

# log scale is often an effective way to handle outliers, especially for features with skewed distributions.
logscaled_df = np.log2(df + 0.01)
logscaled_df.head()

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE
CUST_ID
C10001	5.354408	-0.271980	6.576069	-6.643856	6.576069	-6.643856	-2.500896	-6.643856	-3.421469	-6.643856	-6.643856	1.007196	9.965799	7.656869	7.124326	-6.643856	3.586164
C10002	11.644973	-0.121720	-6.643856	-6.643856	-6.643856	12.653507	-6.643856	-6.643856	-6.643856	-1.943416	2.003602	-6.643856	12.773141	12.002478	10.066560	-2.106423	3.586164
C10003	11.284916	0.014355	9.594661	9.594661	-6.643856	-6.643856	0.014355	0.014355	-6.643856	-6.643856	-6.643856	3.586164	12.872677	9.280949	9.293000	-6.643856	3.586164
C10004	10.702762	-0.629581	10.549794	10.549794	-6.643856	7.685085	-3.421469	-3.421469	-6.643856	-3.421469	0.014355	0.014355	12.872677	-6.643856	-6.643856	-6.643856	3.586164
C10005	9.675471	0.014355	4.000901	4.000901	-6.643856	-6.643856	-3.421469	-3.421469	-6.643856	-6.643856	-6.643856	0.014355	10.228831	9.405875	7.935467	-6.643856	3.586164

minmax_scaler = MinMaxScaler().fit_transform(logscaled_df)

Exploratory Data Analysis (EDA)

minmax_scaler_df = pd.DataFrame(minmax_scaler, columns=df.columns) # Convert the NumPy array to a pandas DataFrame to use .iloc
fig, axs = plt.subplots(nrows=5, ncols=4, figsize = (20,20))

z = 0
for i in range(5):
    for j in range(4):
        if z > 16:
            axs[i][j].axis(False)
        else:
            axs[i][j].set_title(minmax_scaler_df.iloc[:,z].name)
            axs[i][j].hist(minmax_scaler_df.iloc[:,z])
        z+=1

# Create the boxplot
plt.figure(figsize=(15, 8))
sns.boxplot(data=minmax_scaler)
plt.xticks(rotation=90) # Rotate x-axis labels for better readability
plt.title('Boxplot of Features after MinMax Scaling')
plt.show()

# Heatmap
plt.figure(figsize=(12, 10))
sns.heatmap(minmax_scaler_df.corr(), annot=True, cmap='coolwarm', fmt=".2f")
plt.title('Correlation Heatmap of Credit Card Usage Features')
plt.show()

minmax_scaler_df.columns

Index(['BALANCE', 'BALANCE_FREQUENCY', 'PURCHASES', 'ONEOFF_PURCHASES',
       'INSTALLMENTS_PURCHASES', 'CASH_ADVANCE', 'PURCHASES_FREQUENCY',
       'ONEOFF_PURCHASES_FREQUENCY', 'PURCHASES_INSTALLMENTS_FREQUENCY',
       'CASH_ADVANCE_FREQUENCY', 'CASH_ADVANCE_TRX', 'PURCHASES_TRX',
       'CREDIT_LIMIT', 'PAYMENTS', 'MINIMUM_PAYMENTS', 'PRC_FULL_PAYMENT',
       'TENURE'],
      dtype='object')

Heatmap shows multiple features are corelated. apply PCA to get uncorrelated variables

Dimensionality Reduction using Principal Component Analysis

#PCA
pca = PCA(n_components = 17)
pca.fit(minmax_scaler_df)
pca_df = pca.transform(minmax_scaler_df)

#Determine the optimal number of Principal Components

var_ratio = {}
for n in range(0, 17):
    pc = PCA(n_components = n)
    df_pca = pc.fit(minmax_scaler_df)
    var_ratio[n] = sum(df_pca.explained_variance_ratio_*100)

var_ratio

{0: 0,
 1: 49.72413525043054,
 2: 67.74056946986803,
 3: 81.54447914140762,
 4: 89.19895807543327,
 5: 92.43618740550028,
 6: 94.6687108232442,
 7: 96.7517397421888,
 8: 97.70870080693823,
 9: 98.59168333702078,
 10: 99.05986879820107,
 11: 99.3030009153706,
 12: 99.5217458015324,
 13: 99.70170118889953,
 14: 99.83107720527848,
 15: 99.90992497173231,
 16: 99.95648167876995}

pd.Series(var_ratio).plot()

<Axes: >

#Scree plot function
def scree_plot(pca, scaler):
    per_var = np.round(pca.explained_variance_ratio_* 100, decimals=1)
    labels = ['PC' + str(x) for x in range(1, len(per_var)+1)]

    plt.figure(figsize = (12,7))
    plt.bar(x = range(1,len(per_var)+1), height = per_var, tick_label = labels)
    plt.plot(range(1, len(per_var)+1), per_var, c = 'red')
    plt.ylabel('Percentage of Explained Variance', size = 13)
    plt.xlabel('Principal Components', size = 13)
    plt.title('Scree Plot (%s)' %(scaler), size = 15)
    plt.grid()
    plt.show()
    print('%s%% variance of the dataset lies on the first 3 Principal Components'  %(np.sum(per_var[:3])))

scree_plot(pca, 'minmax_scaler')

81.5% variance of the dataset lies on the first 3 Principal Components

#Use the 3 Principal Components
top3_pc = pca_df[:,:3]

#2D visualization of the 3 Principal Components
x = top3_pc[:, 0]
y = top3_pc[:, 1]
z = top3_pc[:, 2]

plt.figure(figsize=(15, 5))

plt.subplot(1, 3, 1)
plt.scatter(x, y)
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.title('PC1 vs PC2')

plt.subplot(1, 3, 2)
plt.scatter(x, z)
plt.xlabel('PC1')
plt.ylabel('PC3')
plt.title('PC1 vs PC3')

plt.subplot(1, 3, 3)
plt.scatter(y, z)
plt.xlabel('PC2')
plt.ylabel('PC3')
plt.title('PC2 vs PC3')

plt.tight_layout()
plt.show()

#Combining the 3 Principal Components to 1 scatter plot

plt.figure(figsize=(10, 8))

plt.scatter(x, y, c=z, cmap='viridis')  # Color code points by z-values
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.title('PC1 vs PC2 (color-coded by PC3)')
plt.colorbar(label='PC3') # Add colorbar for better understanding of the color mapping


plt.show()

#3D scatter
def scatter_plot3D(pca_df):
    x, y, z = pca_df[:,:3].T
    fig = px.scatter_3d(x = x,
                        y = y,
                        z = z)
    fig.show()

scatter_plot3D(pca_df)

Select Clustering Model and Algorithm

Given that we want to capture customer behavior patterns, clustering methods like K-Means, DBSCAN, or Hierarchical Clustering are ideal choices.

#KMeans Clustering
#For value of K, inertia and silhouette score is used.

inertia = []
silhouette_kmeans = []


for i in range(2,10):
  kmeans = KMeans(n_clusters = i)
  kmeans.fit(top3_pc)
  inertia.append(kmeans.inertia_)
  silhouette_kmeans.append(silhouette_score(top3_pc, kmeans.labels_))


#inertia score figure
plt.figure(figsize = (8,4))
plt.plot(range(2,10), inertia)
plt.title('Inertia Score')
plt.xticks(range(1,10))
plt.grid()
plt.show()

Elbow Criterion Method: 7 clusters

this method run k-means clustering on a given dataset for a range of values of k (e.g k=1 to 10), for each value of k, calculate sum of squared errors (SSE).

#Silhouette score figure
plt.figure(figsize = (8,4))
plt.plot(range(2,10), silhouette_kmeans)
plt.title('Silhouette Score')
plt.xticks(range(1,10))
plt.grid()
plt.show()

Silhouette Coefficient Method: 7 clusters.

A higher Silhouette Coefficient score relates to a model with better-defined clusters.

Conclusion: 7 Clusters.

because the elbow is lie at 7th cluster in the inertia score, and the highest value in Silhouette Score is also located at 7th cluster

#KMeans Clustering
kmeans = KMeans(n_clusters=7)
kmeans.fit(top3_pc)
kmeans_labels = kmeans.labels_

#2D visualization of the 3 Principal Components
x = top3_pc[:, 0]
y = top3_pc[:, 1]
z = top3_pc[:, 2]

plt.figure(figsize=(15, 5))

plt.subplot(1, 3, 1)
plt.scatter(x, y, c=kmeans_labels)
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.title('PC1 vs PC2')

plt.subplot(1, 3, 2)
plt.scatter(x, z, c=kmeans_labels)
plt.xlabel('PC1')
plt.ylabel('PC3')
plt.title('PC1 vs PC3')

plt.subplot(1, 3, 3)
plt.scatter(y, z, c=kmeans_labels)
plt.xlabel('PC2')
plt.ylabel('PC3')
plt.title('PC2 vs PC3')

plt.tight_layout()
plt.show()

#3D visualization
x, y, z = top3_pc.T

fig = px.scatter_3d(x = x,
                    y = y,
                    z = z,
                    color = kmeans_labels)
fig.show()

#Agglomerative Clustering
# Finding the best number of clusters using silhouette score

silhouette_agglo = []
for n_clusters in range(2, 10):
    agg_cluster = AgglomerativeClustering(n_clusters=n_clusters)
    agg_cluster.fit(top3_pc)
    silhouette_agglo.append(silhouette_score(top3_pc, agg_cluster.labels_))

# Plotting the silhouette scores
plt.figure(figsize=(8, 4))
plt.plot(range(2, 10), silhouette_agglo)
plt.title('Silhouette Score for Agglomerative Clustering')
plt.xlabel('Number of Clusters')
plt.ylabel('Silhouette Score')
plt.xticks(range(2, 10))
plt.grid()
plt.show()

# Agglomerative Clustering with the optimal number of clusters (7 clusters)
agg_cluster = AgglomerativeClustering(n_clusters=7)
agg_cluster.fit(top3_pc)
agg_labels = agg_cluster.labels_

# 2D visualization of the 3 Principal Components with Agglomerative Clustering labels
x = top3_pc[:, 0]
y = top3_pc[:, 1]
z = top3_pc[:, 2]

plt.figure(figsize=(15, 5))

plt.subplot(1, 3, 1)
plt.scatter(x, y, c=agg_labels)
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.title('PC1 vs PC2 - Agglomerative Clustering')

plt.subplot(1, 3, 2)
plt.scatter(x, z, c=agg_labels)
plt.xlabel('PC1')
plt.ylabel('PC3')
plt.title('PC1 vs PC3 - Agglomerative Clustering')

plt.subplot(1, 3, 3)
plt.scatter(y, z, c=agg_labels)
plt.xlabel('PC2')
plt.ylabel('PC3')
plt.title('PC2 vs PC3 - Agglomerative Clustering')

plt.tight_layout()
plt.show()

# 3D visualization with Agglomerative Clustering labels
x, y, z = top3_pc.T

fig = px.scatter_3d(x=x,
                    y=y,
                    z=z,
                    color=agg_labels,
                    title='3D Scatter Plot - Agglomerative Clustering')
fig.show()

#DBSCAN Clustering
dbscan = DBSCAN(eps = 0.2) # eps = 0.2 is achieved iteratively to get the most reasonable number of clusters.
dbscan.fit(top3_pc)
dbscan_labels = dbscan.labels_ # label/cluster setiap instance

# Visualize the results (similar to KMeans visualization code)
x = top3_pc[:, 0]
y = top3_pc[:, 1]
z = top3_pc[:, 2]

plt.figure(figsize=(15, 5))

plt.subplot(1, 3, 1)
plt.scatter(x, y, c=dbscan_labels)
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.title('PC1 vs PC2 - DBSCAN')

plt.subplot(1, 3, 2)
plt.scatter(x, z, c=dbscan_labels)
plt.xlabel('PC1')
plt.ylabel('PC3')
plt.title('PC1 vs PC3 - DBSCAN')

plt.subplot(1, 3, 3)
plt.scatter(y, z, c=dbscan_labels)
plt.xlabel('PC2')
plt.ylabel('PC3')
plt.title('PC2 vs PC3 - DBSCAN')

plt.tight_layout()
plt.show()

# 3D visualization
x, y, z = top3_pc.T

fig = px.scatter_3d(x=x,
                    y=y,
                    z=z,
                    color=dbscan_labels,
                    title='3D Scatter Plot - DBSCAN')
fig.show()

Evaluate clustering quality using silhouette score

# Calculate silhouette scores for each clustering method
kmeans_silhouette = silhouette_score(top3_pc, kmeans_labels)
agg_silhouette = silhouette_score(top3_pc, agg_labels)
dbscan_silhouette = silhouette_score(top3_pc, dbscan_labels)

print(f"KMeans Silhouette Score: {kmeans_silhouette}")
print(f"Agglomerative Clustering Silhouette Score: {agg_silhouette}")
print(f"DBSCAN Silhouette Score: {dbscan_silhouette}")

# Determine the best clustering method based on the silhouette score
best_method = max([(kmeans_silhouette, "KMeans"), (agg_silhouette, "Agglomerative"), (dbscan_silhouette, "DBSCAN")], key=lambda x: x[0])

print(f"\nBest Clustering Method: \n{best_method[1]} with Silhouette Score: {best_method[0]}")

KMeans Silhouette Score: 0.686522893798625
Agglomerative Clustering Silhouette Score: 0.6858178316373196
DBSCAN Silhouette Score: 0.6833543686053595

Best Clustering Method: 
KMeans with Silhouette Score: 0.686522893798625

Based on the silhouette scores obtained, we can draw the following conclusions:

Clustering Method Comparison:

• KMeans achieved the highest silhouette score of 0.6865, indicating that it provided the most well-separated clusters among the three methods.

• Agglomerative Clustering scored slightly lower at 0.6858, showing that it performed comparably to KMeans but with marginally less separation between clusters.

• DBSCAN had a silhouette score of 0.6834, the lowest of the three. This result suggests that DBSCAN’s clusters were somewhat less cohesive or had more overlap, possibly due to varying density in the data, which DBSCAN is sensitive to.

Best Clustering Method:

Given the highest silhouette score, KMeans is identified as the best clustering method for this dataset. Its high score (close to 1) suggests strong, distinct clustering with minimal overlap, making it an effective choice for this segmentation task.

Implications for Customer Segmentation:

KMeans’ clear and slightly higher score indicates it will likely provide the most useful segmentation for understanding customer behaviors in this dataset. This method offers well-defined clusters, which can facilitate more targeted marketing, accurate risk assessment, and improved customer relationship strategies.

In conclusion, KMeans is the recommended clustering model for segmenting credit card customers based on their spending and payment behaviors. Its strong silhouette score reflects well-separated, cohesive clusters, which will be valuable for practical application in customer segmentation efforts.

Interpret and Analyze Cluster Profiles

#Calculate Cluster Statistics
labels, counts = np.unique(kmeans_labels, return_counts = True)

plt.figure(figsize = (8,6))
plt.title('Quantity of Each Cluster', size = 16)
plt.bar(labels, counts)
plt.xlabel('Cluster', size = 14)
plt.ylabel('Customers', size = 14)
plt.xticks(labels)
plt.show()

for i in zip(labels, counts):
    print('Cluster %s: %s' %(i[0],i[1]))

Cluster 0: 1740
Cluster 1: 2045
Cluster 2: 1804
Cluster 3: 804
Cluster 4: 1084
Cluster 5: 1012
Cluster 6: 460

#create a file with the label
df_final = df.copy()
df_final['Label'] = kmeans_labels
df_final.to_csv('final_data.csv') # save final dataset for next step, interpretation, conclusion, and recommendation
df_final.head()

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
CUST_ID
C10001	40.900749	0.818182	95.40	0.00	95.4	0.000000	0.166667	0.000000	0.083333	0.000000	0	2	1000.0	201.802084	139.509787	0.000000	12	2
C10002	3202.467416	0.909091	0.00	0.00	0.0	6442.945483	0.000000	0.000000	0.000000	0.250000	4	0	7000.0	4103.032597	1072.340217	0.222222	12	1
C10003	2495.148862	1.000000	773.17	773.17	0.0	0.000000	1.000000	1.000000	0.000000	0.000000	0	12	7500.0	622.066742	627.284787	0.000000	12	4
C10004	1666.670542	0.636364	1499.00	1499.00	0.0	205.788017	0.083333	0.083333	0.000000	0.083333	1	1	7500.0	0.000000	0.000000	0.000000	12	3
C10005	817.714335	1.000000	16.00	16.00	0.0	0.000000	0.083333	0.083333	0.000000	0.000000	0	1	1200.0	678.334763	244.791237	0.000000	12	4

#Describe each cluster
for i in range(7):
    print('Cluster:', i)
    display(df_final[df_final['Label'] == i].describe())
    print('\n')

Cluster: 0

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	1740.000000	1740.000000	1740.000000	1740.000000	1740.000000	1740.0	1740.000000	1740.000000	1740.000000	1740.0	1740.0	1740.000000	1740.000000	1740.000000	1740.000000	1740.000000	1740.000000	1740.0
mean	1220.881525	0.935343	2484.258443	1519.791874	964.494730	0.0	0.816874	0.463385	0.649881	0.0	0.0	34.704023	5713.466562	2293.289675	610.868334	0.257684	11.807471	0.0
std	1819.737542	0.156607	3501.511631	2786.850472	1338.061998	0.0	0.232588	0.332811	0.310346	0.0	0.0	36.533021	4001.755686	3256.661299	1624.248410	0.357874	0.829482	0.0
min	0.000000	0.000000	34.000000	1.400000	1.950000	0.0	0.083333	0.083333	0.083333	0.0	0.0	2.000000	300.000000	0.000000	0.000000	0.000000	6.000000	0.0
25%	167.319455	1.000000	749.825000	271.630000	232.922500	0.0	0.666667	0.166667	0.416667	0.0	0.0	13.000000	2500.000000	698.994832	169.446775	0.000000	12.000000	0.0
50%	533.733514	1.000000	1538.185000	773.660000	546.195000	0.0	0.916667	0.408334	0.666667	0.0	0.0	24.000000	5000.000000	1345.306081	212.079018	0.000000	12.000000	0.0
75%	1518.458175	1.000000	2856.992500	1752.042500	1176.730000	0.0	1.000000	0.750000	1.000000	0.0	0.0	42.000000	7500.000000	2602.843109	549.842969	0.500000	12.000000	0.0
max	19043.138560	1.000000	41050.400000	40624.060000	15497.190000	0.0	1.000000	1.000000	1.000000	0.0	0.0	358.000000	30000.000000	36066.750680	29019.802880	1.000000	12.000000	0.0

Cluster: 1

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.000000	2045.0
mean	2153.675875	0.882876	0.000034	0.000034	0.042518	1994.138477	0.000126	0.000126	0.000122	0.272754	6.304645	0.003423	4031.426243	1655.244249	972.278185	0.044363	11.321271	1.0
std	2119.873884	0.231895	0.001149	0.001149	1.544925	2472.248061	0.003290	0.003290	0.004120	0.201041	8.078662	0.066268	3269.443304	2715.813753	2574.175018	0.138180	1.544184	0.0
min	0.000000	0.000000	0.000000	0.000000	0.000000	18.042768	0.000000	0.000000	0.000000	0.083333	1.000000	0.000000	50.000000	0.000000	0.000000	0.000000	6.000000	1.0
25%	753.703481	0.900000	0.000000	0.000000	0.000000	380.852744	0.000000	0.000000	0.000000	0.090909	2.000000	0.000000	1500.000000	360.416137	249.298802	0.000000	12.000000	1.0
50%	1455.213589	1.000000	0.000000	0.000000	0.000000	1209.369868	0.000000	0.000000	0.000000	0.250000	4.000000	0.000000	3000.000000	764.938728	471.885081	0.000000	12.000000	1.0
75%	2843.942194	1.000000	0.000000	0.000000	0.000000	2706.493856	0.000000	0.000000	0.000000	0.333333	8.000000	0.000000	6000.000000	1730.364781	983.378772	0.000000	12.000000	1.0
max	14581.459140	1.000000	0.050000	0.050000	66.950000	26194.049540	0.090909	0.090909	0.166667	1.500000	123.000000	2.000000	19000.000000	34107.074990	61031.618600	1.000000	12.000000	1.0

Cluster: 2

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	1804.000000	1804.000000	1804.000000	1804.000000	1804.000000	1804.0	1804.000000	1804.000000	1804.000000	1804.0	1804.0	1804.000000	1804.000000	1804.000000	1804.000000	1804.000000	1804.000000	1804.0
mean	398.592191	0.787668	537.079074	0.018315	538.067450	0.0	0.705528	0.000185	0.671907	0.0	0.0	11.805432	3121.138239	748.728568	680.650554	0.306385	11.509424	2.0
std	916.359635	0.300094	886.586469	0.776954	888.588269	0.0	0.306414	0.003921	0.309206	0.0	0.0	12.180076	2712.773792	1329.716714	3243.687768	0.368776	1.372586	0.0
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.0	0.083333	0.000000	0.083333	0.0	0.0	0.000000	300.000000	0.000000	0.000000	0.000000	6.000000	2.0
25%	18.270355	0.636364	154.965000	0.000000	154.965000	0.0	0.416667	0.000000	0.416667	0.0	0.0	6.000000	1400.000000	208.655774	108.813746	0.000000	12.000000	2.0
50%	56.075919	1.000000	321.360000	0.000000	322.145000	0.0	0.833333	0.000000	0.750000	0.0	0.0	10.000000	2500.000000	419.418415	163.036039	0.095454	12.000000	2.0
75%	288.875684	1.000000	601.897500	0.000000	604.620000	0.0	1.000000	0.000000	1.000000	0.0	0.0	12.000000	4000.000000	825.695754	217.010143	0.600000	12.000000	2.0
max	11670.179850	1.000000	22500.000000	33.000000	22500.000000	0.0	1.000000	0.083333	1.000000	0.0	0.0	199.000000	23000.000000	28988.973880	76406.207520	1.000000	12.000000	2.0

Cluster: 3

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.000000	804.0
mean	2332.116133	0.923417	661.071530	660.831679	0.283035	2015.911075	0.275573	0.275366	0.000489	0.292519	6.820896	6.287313	4622.667345	1910.424994	991.348672	0.057587	11.348259	3.0
std	2271.592390	0.174668	1497.048804	1497.096309	4.538610	2476.547218	0.252572	0.252696	0.007150	0.212246	7.727634	13.487905	3495.941761	2906.525630	1664.230553	0.165666	1.521325	0.0
min	0.000000	0.000000	0.700000	0.700000	0.000000	18.280043	0.083333	0.083333	0.000000	0.083333	1.000000	1.000000	200.000000	0.000000	0.000000	0.000000	6.000000	3.0
25%	816.313793	1.000000	114.155000	114.155000	0.000000	367.425769	0.083333	0.083333	0.000000	0.142857	2.000000	1.000000	1987.500000	480.378702	244.062670	0.000000	12.000000	3.0
50%	1600.350201	1.000000	296.500000	296.500000	0.000000	1138.401787	0.166667	0.166667	0.000000	0.250000	4.000000	3.000000	3450.000000	1067.160999	533.069335	0.000000	12.000000	3.0
75%	3116.558155	1.000000	720.600000	720.600000	0.000000	2745.786312	0.333333	0.333333	0.000000	0.416667	9.000000	7.000000	6500.000000	2185.613015	1144.611880	0.000000	12.000000	3.0
max	15258.225900	1.000000	22101.780000	22101.780000	103.880000	20277.331120	1.000000	1.000000	0.142857	1.250000	93.000000	309.000000	20000.000000	39461.965800	19060.534750	1.000000	12.000000	3.0

Cluster: 4

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	1084.000000	1084.000000	1084.000000	1084.000000	1084.000000	1084.0	1084.000000	1084.000000	1084.000000	1084.0	1084.0	1084.000000	1084.000000	1084.000000	1084.000000	1084.000000	1084.000000	1084.0
mean	761.385342	0.768659	874.372749	874.007897	0.451504	0.0	0.352508	0.351790	0.001109	0.0	0.0	7.677122	4422.841888	1191.039662	476.966500	0.134404	11.553506	4.0
std	1178.933402	0.304491	1398.825474	1399.025745	5.048241	0.0	0.312450	0.312835	0.010372	0.0	0.0	13.739261	3614.781349	2336.538055	1590.631835	0.291630	1.278379	0.0
min	0.000000	0.000000	0.010000	0.010000	0.000000	0.0	0.083333	0.083333	0.000000	0.0	0.0	0.000000	150.000000	0.000000	0.000000	0.000000	6.000000	4.0
25%	44.087865	0.545455	150.000000	149.887500	0.000000	0.0	0.083333	0.083333	0.000000	0.0	0.0	1.000000	1600.000000	285.646717	110.465810	0.000000	12.000000	4.0
50%	245.268294	1.000000	437.805000	435.940000	0.000000	0.0	0.236111	0.222222	0.000000	0.0	0.0	4.000000	3000.000000	572.172550	188.098563	0.000000	12.000000	4.0
75%	1089.196875	1.000000	1023.080000	1023.080000	0.000000	0.0	0.500000	0.500000	0.000000	0.0	0.0	10.000000	6500.000000	1273.133283	410.594686	0.083333	12.000000	4.0
max	12323.845360	1.000000	17945.000000	17945.000000	106.970000	0.0	1.000000	1.000000	0.166667	0.0	0.0	217.000000	25000.000000	50721.483360	30528.432400	1.000000	12.000000	4.0

Cluster: 5

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.000000	1012.0
mean	2838.387763	0.970120	1938.609822	1165.128715	773.784634	2091.708367	0.780558	0.397527	0.630515	0.292225	7.750000	30.568182	5819.675011	3023.259338	1237.123901	0.069474	11.619565	5.0
std	2617.941642	0.089857	2535.797505	1938.815928	1036.273615	3109.109798	0.244183	0.289533	0.308479	0.213466	9.740945	29.750467	4139.102367	4088.312470	1767.758520	0.176262	1.188692	0.0
min	2.299436	0.181818	64.820000	0.020000	9.650000	14.222216	0.083333	0.083333	0.083333	0.083333	1.000000	2.000000	300.000000	0.000000	0.000000	0.000000	6.000000	5.0
25%	898.144287	1.000000	612.172500	227.065000	186.502500	397.425614	0.583333	0.166667	0.375000	0.090909	2.000000	11.000000	2500.000000	879.043044	328.232419	0.000000	12.000000	5.0
50%	2014.049593	1.000000	1243.580000	616.625000	447.230000	1130.336826	0.875000	0.333333	0.666667	0.250000	4.500000	21.000000	5000.000000	1747.866882	741.987527	0.000000	12.000000	5.0
75%	4017.065867	1.000000	2394.680000	1407.717500	1021.405000	2650.504279	1.000000	0.583333	0.916667	0.416667	10.000000	39.000000	8000.000000	3539.576179	1452.988501	0.083333	12.000000	5.0
max	16259.448570	1.000000	49039.570000	40761.250000	13184.430000	47137.211760	1.000000	1.000000	1.000000	1.000000	123.000000	308.000000	30000.000000	46930.598240	22011.779800	1.000000	12.000000	5.0

Cluster: 6

	BALANCE	BALANCE_FREQUENCY	PURCHASES	ONEOFF_PURCHASES	INSTALLMENTS_PURCHASES	CASH_ADVANCE	PURCHASES_FREQUENCY	ONEOFF_PURCHASES_FREQUENCY	PURCHASES_INSTALLMENTS_FREQUENCY	CASH_ADVANCE_FREQUENCY	CASH_ADVANCE_TRX	PURCHASES_TRX	CREDIT_LIMIT	PAYMENTS	MINIMUM_PAYMENTS	PRC_FULL_PAYMENT	TENURE	Label
count	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.000000	460.0
mean	2568.605387	0.956652	534.756826	0.000022	535.406348	2054.554450	0.653573	0.000181	0.606247	0.262339	6.208696	12.286957	4356.739130	1954.383558	1431.422757	0.047027	11.319565	6.0
std	2594.263464	0.125205	847.334096	0.000466	847.252725	2545.640406	0.313629	0.003885	0.314303	0.197822	8.951815	19.504992	3614.767613	3076.065407	2524.702344	0.128253	1.614223	0.0
min	1.097736	0.181818	4.440000	0.000000	4.440000	18.531069	0.083333	0.000000	0.083333	0.083333	1.000000	1.000000	500.000000	0.000000	0.000000	0.000000	6.000000	6.0
25%	879.346363	1.000000	158.067500	0.000000	158.067500	507.772247	0.356060	0.000000	0.333333	0.083333	2.000000	5.000000	1500.000000	577.127243	344.167441	0.000000	12.000000	6.0
50%	1629.791663	1.000000	335.995000	0.000000	336.990000	1133.902290	0.707143	0.000000	0.666667	0.166667	4.000000	9.000000	3000.000000	1131.976678	798.641580	0.000000	12.000000	6.0
75%	3344.281991	1.000000	625.670000	0.000000	625.670000	2682.999427	1.000000	0.000000	0.909091	0.333333	7.000000	13.000000	6000.000000	2052.086325	1573.285404	0.000000	12.000000	6.0
max	16304.889250	1.000000	14686.100000	0.010000	14686.100000	19862.878440	1.000000	0.083333	1.000000	1.000000	123.000000	347.000000	19000.000000	40627.595240	31871.363790	1.000000	12.000000	6.0

#Violin Plots
fig, axs = plt.subplots(nrows = 17, ncols = 1, figsize = (16,70))
for i in range(17):
    q1, q3 = np.quantile(df_final.iloc[:,i], [0.25, 0.75])
    sns.violinplot(data = df_final[['Label', df_final.columns[i]]].loc[df_final[df_final.columns[i]] <= (q3 + 1.5*(q3-q1))], x = 'Label', y = df_final.columns[i], ax = axs[i])

plt.show()

Cluster Insights and recommendations:

Cluster 0: Moderate Spenders with Stable Usage

This cluster likely represents average spenders with moderate credit use and lower dependency on cash advances, suggesting a stable and reliable customer group.

• Loyalty Programs: Offer points-based or cashback rewards to encourage continued usage and build loyalty.
• Cross-Sell Opportunities: Introduce additional financial products like personal loans or savings accounts that may appeal to moderate spenders.

Cluster 1: High Credit Limit & Low Usage

This group may represent inactive or conservative users with high credit limits but minimal spending. They may need engagement to increase activity or represent low-risk, low-revenue customers.

• Re-engagement Campaign: Offer incentives like cashback on the first purchase or waiver of annual fees to encourage card usage.
• Usage-Based Promotions: Create targeted campaigns to incentivize specific spending categories, such as travel or dining, which may appeal to more conservative or inactive users.

Cluster 2: Low Income / Newer Users with Low Balances and Credit Limits

This cluster likely represents new or low-income users who have lower balances and credit limits. They may benefit from incentives to increase engagement or credit limits.

• Introductory Offers: Provide promotions, like introductory 0% interest on purchases, to help these newer customers get accustomed to using their credit cards.
• Gradual Credit Limit Increases: Based on payment history, consider offering small, incremental credit limit increases to build customer trust and engagement.

Cluster 3: High-Spending, High-Engagement Customers with Cash Advance Dependency

This cluster likely represents high-value, high-spending customers with high credit limits. However, their frequent use of cash advances could indicate liquidity needs, and they may need monitoring for credit risk or engagement for premium services.

• ** Premium Credit Card Options:** Offer upgrades to premium cards with additional perks like travel rewards, concierge services, or access to exclusive events to deepen engagement.

Cluster 4: Mid-Tier Stable Spenders with Low Cash Advance Dependency

This cluster could represent middle-income customers with stable spending patterns and little need for cash advances. They may be prime candidates for targeted offers to increase loyalty.

• Upgrade Incentives: Encourage these stable users to upgrade to a higher-tier card by offering benefits like rewards points, annual fee waivers, or travel-related perks.
• Increase Engagement: Target them with campaigns for higher spend categories, like travel or luxury, or incentivize certain types of purchases (e.g., dining, entertainment) to increase card usage.

Cluster 5: Long-Term, Low-Spending Customers with Low Balances

This cluster represents low-engagement, low-risk customers who do not spend much or use their credit cards actively. They may benefit from reactivation campaigns or low-cost service offers to boost engagement.

• Retention Strategies: Offer them fee waivers or basic loyalty points to keep them engaged and prevent attrition, as they are likely low-risk but also low-revenue customers.
• Simple Upgrade Offers: Provide low-cost upgrade options that give modest rewards, appealing to conservative spenders.

Cluster 6: High-Value, High-Spending Customers with Frequent Cash Advances

This cluster represents a high-value segment with very high spending and credit use, including frequent cash advances. They are prime candidates for premium services and loyalty rewards but may need careful monitoring due to high credit usage.

• Exclusive Benefits: Offer highly personalized rewards and services, like concierge services, airport lounge access, or exclusive event invitations. Consider assigning a dedicated relationship manager.
• Financial Management Services: Given the frequent use of cash advances, they may benefit from specialized financial management services. Offer advisory services or access to personal loans as an alternative to cash advances.

Summary of Recommendations

Cluster 0 & 4: Mid-tier customers, stable users who may respond well to loyalty rewards.

Cluster 1 & 5: Low-spending or inactive users, candidates for re-engagement campaigns.

Cluster 2: New or low-limit customers, could benefit from increased credit limits and introductory offers.

Cluster 3 & 6: High-value customers with premium credit usage and cash advance dependency; ideal for premium service offers but require close risk monitoring.

Each cluster’s unique behavior warrants a targeted approach to maximize customer satisfaction, reduce credit risk, and improve overall profitability for the bank. By tailoring offerings and engagement strategies to each segment’s needs, the bank can strengthen customer relationships and optimize its credit card portfolio.